Comparison of existing 3-D velocity models of the Kanto basin in terms of long-period ground motions for the hypothetical Tokai earthquake

Abstract

In this study, we used the 3-D velocity models, proposed by Headquarters for Earthquake Research Promotion (HERP, 2012) and Yamada and Yamanaka (2012), of the Kanto basin for long-period (2~20s) ground motion simulations for the hypothetical Tokai earthquake (Mw~8). We used published source model for the earthquake. Our previous study (Dhakal and Yamanaka, 2013) showed that the velocity models performed nearly similarly in the period range of 2 to 10 s at most sites in the Kanto basin, and differed at relatively fewer sites, for moderate magnitude (Mw<6.0) earthquakes which occurred beneath the basin. In this study, we compared the synthetics in the period range of 2 to 20s between the two models for the large magnitude Tokai earthquake. We found that the two models generate similar velocity waveforms and their spectra at most sites in the Kanto basin. To describe the similarity between the synthetics quantitatively, we obtained a goodness of fit between the two synthetics based on three parameters, PGVs, smoothed Fourier spectra (FFT), and cross correlations, using an algorithm proposed by Olsen and Mayhew (2010). We found that the fit belongs to mostly fair fit class and higher. This suggests that for the anticipated Tokai earthquake, the two models predict nearly similarly in the period range considered in this study. We found some differences between the two models at a few sites, where the two models have also performed differently for the moderate magnitude events. These results also suggest that a validation of the models based on moderate magnitude earthquakes is useful for the prediction of long-period ground motions for the large magnitude earthquakes.